We seem to have an issue with OpenLDAP not releasing transactions to the
backend bdb database we are using. I've searched as best I can for a
solution to no avail.
Currently we are trying to bulk load 600,000 users to OpenLDAP, 50,000 at
a time. While the load process is running, we are generating a large
number of tranaction logs in bdb. We are trying to run 'db_checkpoint
-1' and 'db_archive -d' in order to trim the transaction logs to a
manageable size as the load runs. However, these do not seem to work
while slapd is running. When we stop slapd, both programs run as
expected and the logs are deleted.
We have tried adding the checkpoint parameter to the slapd.conf, but this
does not seem to work. The docs and entrie in the FAQ-o-matic seem to
indicate that this may not even be supported anymore. There were a couple
of hits on using txn_checkpoint in DB_CONFIG, but this parameter does not
seem to valid.
Is there an issue with checkpointing the backend bdb while slapd is
running? If not, is there a general procedure for forcing checkpoint
without taking slapd down.
We are using OpenLDAP version 2.2.15 on RH ES 30 and bdb 4.2.52.
TIA for any assistance!!